Charisma transform

It's an interesting quirk of our biology that some ways of representing data are much more meaningful than others. We have a very high visual bandwidth, for example, so charts and graphs are much easier for us to understand than numbers. Not so for computers, where the visual information would have to be reverse-engineered back into numbers before it could be analysed. We are similarly very good at understanding movement, but it's trickier to represent things kinematically (though there are some pretty amazing experiments already).

Rather than visualisations or animations, I think of these functional attempts to shift data into a more easily digestible form as transforms, no different to the kind of transform you might do when converting data between file formats. It's just that these formats are tailored for our own peculiar data ingestion engine. You could say that transforms like these are designed to exploit particular capabilities of our hardware.

There's one capability that I think is both powerful and underexplored: our empathy. As inherently social creatures, we are very efficient at understanding and simulating the behaviour of others. However, most tools we understand mechanistically, like a car or a keyboard; you know that everything that happens follows directly from something else. But it only works for simple behaviour. How do you understand the behaviour of a complex network, or a country's economy? Doing a strict mechanistic analysis is too hard in many cases to be useful.

You can understand these complex behaviours much more easily if you can transform them into the empathic domain where we have specialised understanding. And if we could find a way to effectively describe the motivations of an economy, say, by casting the major forces as characters, assigning them emotions and values that reflect their real-world behaviour, I think it would make the whole thing a lot more intuitive.

What's more, our tools are rapidly exceeding the complexity where we can reason about them in anything but an abstract way. Historical computer interfaces had a simple mapping between actions and results; there's a list of commands, you type a command from the list, it does the same thing every time. But what about voice interfaces? Or search results and other name queries? Or really any system with a user model? Smarter computers are, unfortunately, less predictable computers.

I believe that to keep these complex tools usable we will need to develop a charisma transform: something that can represent the behaviour of that tool in a humanlike way that we can more easily model. I think our interfaces will have to develop personalities, or something that we can understand the way we understand a personality. I expect this will take some time and most of the early attempts to be pretty ham-fisted, but it seems inevitable that we'll have to go in this direction as systems become more complex and our capability to logically understand them gives out.